Commonly used frequency-based statistical methods only allow data from previous studies to be factored into new trials at the design stage, rather than the analysis stage where information is considered an adjunct to, not part of, new results.
In contrast, the Bayesian approach being advocated by the FDA view data from both old and new studies as a continuous stream of information, enabling available information to be considered during results analysis.
The FDA believes that: “When good prior information on clinical use of a device exists, the Bayesian approach may enable this information to be incorporated into the statistical analysis of a trial. In some circumstances, the prior information for a device may be a justification for a smaller-sized or shorter-duration pivotal trial.”
According to the agency Bayesian analysis is particularly suited to medical devices because, unlike pharmaceuticals, their effects are usually physical and therefore limited to specific areas rather than systemic.
It added that: “As a result, device effects are typically local, not systemic. Local effects can sometimes be predictable from prior information on the previous generations of a device when modifications to the device are minor.
The FDA went on to says that Bayesian methods can also be used to include information from medical device trials conducted overseas, which would be more problematic with drugs.
In a statement accompanying the launch, FDA commissioner Margaret Hamburg said the new guidelines are consistent with the agency’s “commitment to streamline clinical trials, when possible, in order to get safe and effective products to the market faster.”
While the FDA's support for Bayesian analysis is patient focused, the new guidance focus on cost reduction means will surely be welcomed by medical device developers and the increasing number of contract research organisations (CROs) offering biostatistics services.
The guidance document can be found here.